The American Journal of Managed Care August 2009
Can a Nationwide Media Campaign Affect Antibiotic Use?
A nationwide media campaign aimed at parents was associated with reductions in the use of antibiotics for pediatric upper respiratory infections, otitis media, and pharyngitis.
Objective: To evaluate the effectiveness of a nationwide media campaign to reduce antibiotic overuse among children.
Study Design: Prospective observational study of the pediatric population of a health maintenance organization (HMO) comparing antibiotic use during the baseline (November 2004-February 2005) and study (November 2005-February 2006) periods.
Methods: During January 2006 the HMO conducted a media campaign to increase public awareness of the risks of misusing antibiotics, particularly for influenza-like diseases. Antibiotic purchasing rates during specific periods in the study winter were compared with those during corresponding periods in the baseline winter among children diagnosed with upper respiratory infection (URI), otitis media (OM), or pharyngitis. After the intervention, a random subset of the study population was surveyed by telephone to estimate the level of exposure to the campaign and attitudes toward antibiotic use.
Results: The study population consisted of 101,401 children in the baseline winter and 84,979 in the study winter. We noted reductions in antibiotic purchasing for URI, OM, and pharyngitis during the postintervention period compared with the preintervention period (URI odds ratio [OR] = 0.75, 95% confidence interval [CI] = 0.69, 0.81; OM OR = 0.65, 95% CI = 0.59, 0.72; pharyngitis OR = 0.93, 95% CI = 0.89, 0.97). Parents of children with URI exposed to the media campaign were more likely to agree with standards of appropriate antibiotic use than parents not exposed (F1 = 4.18, P = .04).
Conclusions: A media campaign aimed at changing patient behavior can contribute to reducing the rate of inappropriate antibiotic use.
(Am J Manag Care. 2009;15(8):529-534)
Overuse of antibiotics, particularly for the treatment of viral illnesses for which these medications offer no benefit, contributes to the development of antibiotic resistance and increases in costs. Our study of a media campaign stressing appropriate antibiotic use demonstrated the following:
- A campaign based on television programming was an effective way to impart knowledge regarding appropriate antibiotic use to a large audience.
- Parents exposed to a media campaign demonstrated greater knowledge regarding appropriate antibiotic use than parents not exposed.
- Antibiotic use for presumed viral conditions dropped in the period after the campaign.
Physicians may overprescribe antibiotics as a response to patient expectations.9 It also has been suggested that under the pressure of high caseloads, especially during winter months, physicians lack either the time or the skills necessary to explain why antibiotics have no clinical effect on viral infections9,10 and to offer alternative supportive treatment such as a “wait-and-see” approach for acute OM.11
Effective strategies for reducing the misuse of antibiotics should include physician education as well as an effort to decrease patient demand.12 Such a multifaceted approach is taken by the National Campaign for Appropriate Antibiotic Use of the US Centers for Disease Control and Prevention, which recommends physician and medical student education as well as information for patients.13 Community-wide educational campaigns have been shown to lead to reductions in antibiotic purchasing14-16 and demand for antibiotics on the part of parents.17
Overprescription of antibiotics also is a problem in the Israeli healthcare system. Use of antibiotics to treat viral URI and influenza is an issue of particular concern, because it is estimated that up to 10% of the Israeli population suffers from influenza annually, creating a substantial burden on health services.18 During the winter of 2005-2006 Maccabi Healthcare Services (MHS), Israel’s second largest health maintenance organization (HMO), initiated a nationwide media campaign aimed at increasing awareness of the misuse of antibiotics among the general public. The main focus of the campaign was on the inappropriate use of antibiotics in the treatment of influenza and URI. We report the results of a populationbased study conducted to evaluate the impact of this campaign on parental knowledge and awareness of appropriate antibiotic use as well as on patterns of actual antibiotic use.
Maccabi Healthcare Services is responsible for providing health services for 1.7 million members, approximately 25% of the Israeli population. In January 2006, MHS initiated a comprehensive mass media campaign to increase public awareness of the risk of misusing antibiotics, particularly for the treatment of influenza-like diseases. The campaign, which consisted of radio and television advertisements, in conjunction with a concurrent 4-part television series, was targeted at parents of children. The advertisements projected the general message that antibiotics are not an appropriate treatment for colds and other viral URIs. The television series presented the serious implications of misusing antibiotics. The media campaign was carried out during a 2-week period in January 2006 to affect the peak of antibiotic use in the months of January and February of every year.
Assessments of Campaign Effectiveness
We evaluated the effectiveness of the campaign in 2 ways. First, to determine whether the campaign affected actual use of antibiotics, we conducted an analysis of medical and administrative information drawn from the MHS central computer database. The impact of the campaign on antibiotic use was assessed by comparing monthly antibiotic purchasing rates during the preintervention (November-December 2005) and postintervention (January-February 2006) periods of the study winter with the corresponding periods of the baseline winter (November 2004-February 2005). Second, a random sample of parents of children diagnosed with URI were surveyed by telephone to assess changes in attitudes toward antibiotic use. The survey was intended to assess the level of exposure to the campaign and to allow for a comparison of attitudes toward antibiotic use among respondents exposed and unexposed to the campaign. The research received approval from the MHS research committee.
All systems of care in MHS are computerized and data are channeled to and stored in a central database in which a member identification number is used to link information from different sources. Data from all physician visits are captured and stored in a central computer system. For the periods under study, we identified all physician visits of children under the age of 18 years for whom a diagnosis of URI, pharyngitis, OM, or influenza was recorded. (A list of diagnoses is provided in eAppendix A available at www.ajmc.com.) For each case identified, we retrieved the visit date, diagnosis, patient identification number, and physician identification number. Demographic indicators available from membership files include age, sex, status as an immigrant from the Commonwealth of Independent States after 1990, and residence in Arab villages or neighborhoods with a high percentage of ultraorthodox residents. We have found that recent immigrants and members of minority populations differ in their use of medical services; therefore, we included this demographic information in the analysis.
The study focused on the impact of the campaign on the use of the antibiotics for the treatment of acute viral conditions; therefore we excluded children for whom antibiotic treatment might have been provided for another diagnosis or for chronic conditions. For this reason, children meeting any of the following criteria were excluded from the study population: (1) additional diagnoses recorded during the same physician visit; (2) instances of any diagnoses of URI, OM, or pharyngitis in the 6 months preceding the physician visit; or (3) antibiotic purchase within 30 days prior to the physician visit.
The outcome variable of interest was an antibiotic purchase within 3 days of a first visit for URI, OM, or pharyngitis. For all eligible children in the analysis data set, we searched the MHS pharmacy database for purchases of antibiotics prescribed by the diagnosing physician occurring within 3 days after the patient visit. All antibiotic purchases require physician prescription and are subsidized by the HMO; therefore, we assumed that we would capture the vast majority of purchases. We limited our analysis to purchases within a relatively short period and required that the diagnosing physician be the same as the prescribing physician to maximize the likelihood that the diagnosis and antibiotic purchase were indeed linked.
The telephone survey commenced several days after the campaign officially ended and was conducted over a period of 2 weeks. The survey sample consisted of 860 randomly selected parents of MHS members identified in the course of the antibiotic utilization analysis described above. We selected the survey population from among those with URI as it was the most common condition studied, accounting for 57.1% of the study population in the baseline year. Of those contacted, 53% (n = 456) completed interviews.
Two questions were used to assess the respondents’ level of exposure to the campaign. Respondents were first asked, “Do you remember seeing or hearing a recent advertisement about treatment by antibiotics?” Afterward, full descriptions of the advertisements on television and radio were presented, and the respondents were asked to recall if they had seen or heard the advertisements. Those who answered yes to at least 1 of the questions were classified as “exposed.” Those who answered no to both questions were classified as “not exposed.”
Parental knowledge regarding correct use of antibiotics was assessed using a series of statements (see eAppendix B available at www.ajmc.com.) For each statement the respondent was asked to quantify his or her level of agreement, using a scale ranging from 1 (“strongly disagree”) to 10 (“strongly agree”). Each statement reflected an aspect of knowledge regarding appropriate antibiotic use. For example: “Use of unnecessary antibiotics can result in the development of antibiotic-resistant bacteria.” We generated a composite variable, counting the number of responses of 9 or 10 (high level of agreement) for each respondent. Thus, a high score indicated strong agreement with standards for appropriate use of antibiotics, whereas a low score indicated poor agreement. We excluded from the analysis those items to which more than 20% of respondents answered “don’t know.”
To assess the effectiveness of the campaign in reducing rates of antibiotic purchases, the study winter was divided into 2 parts: the 2 months prior to the intervention (November and December) and the 2 months during and after the intervention (January and February). We used a series of binary logistic regression models adjusted for demographic factors to compare rates of antibiotic purchase in the preintervention and postintervention periods of the study winter with the parallel periods in the preceding winter. Binary logistic regression models were constructed for each of the diagnoses of interest, using a Bonferroni correction to hold the overall P value for the model to .05. For each of the 3 conditions studied, we constructed 2 regression models to compare antibiotic purchasing in the baseline winter versus the study winter. The first model in each pair focused on the preintervention period and the second focused on the postintervention period. In addition, we used general linear models to estimate the effect of exposure to the campaign, as determined by survey, with the level of agreement with appropriate use of antibiotics. All analyses were performed using SPSS version 14 (SPSS Inc, Chicago, IL).
During the baseline winter, 206,235 children under age 18 years received any of the diagnoses of interest. Of these, 101,401 children had no instances of these diagnoses within the preceding 6 months, had no other diagnoses recorded at the same physician visit, and had not received antibiotics in the preceding 30 days, and therefore met the criteria for inclusion in the study population. The corresponding figures for the study winter were 180,110 and 84,979, respectively. During the study winter, URI was the most commonly identified condition in the population under study (53.2%), followed by pharyngitis (39.2%). Otitis media accounted for an additional 7.6% of children (Table 1). The small number of cases of influenza disease (1073 patients during the study winter) were excluded from analysis.